Closed ZisongXu closed 6 months ago
@ZisongXu It happened with me too when I was working with higher resolution images. The initial pose estimation algorithm takes a lot of GPU but only for a couple of seconds and then it drops. So try resizing the rgb, depth and masks to a smaller resolution and then run. Make sure to change cam_K.txt by the same downscale factor while you resize them. That should help you run it on your machine. I am not sure if there is any other way around it. @wenbowen123 can suggest something maybe
@ZisongXu how much GPU memory do you have?
@wenbowen123 My GPU is 3080, memory is 10G I think
https://github.com/NVlabs/FoundationPose/blob/7b484cf03436b9bb6038768e6edd4bc06d759513/run_demo.py#L44
You can set shorter_side
to something smaller, e.g. 480
Nice it works! Thanks a lot!!
Thanks for the nice work. When I tried mustard0, and it works fine. But when I try to run other objects such as driller (example in your link) and cheezit (collected by myself), it had this error:
I want to ask is that really because CUDA out of memory, or because I did something wrong when I change objects.
By the way, if I want to run some experiments use my own data, what should I do? I only changed two lines in run_demo.py script:
Best Regards Zisong XU